Three-dimensional Oil Well Planning in Ultra-deep Water

1Paulo G. de Barros,

1Dinaldo A. Pessoa,

1Pedro J. S. Leite,

1Rodrigo C. Farias,

2Veronica Teichrieb,

1Judith Kelner

1Centro de Informática, Universidade Federal de Pernambuco

2Departamento de Sistemas Computacionais, Escola Politécnica de Pernambuco,

Universidade de Pernambuco

{pgb, dap2, pjsl, rcf3, jk}@cin.ufpe.br, vt@dsc.upe.br

Abstract. This paper presents the Vis-Petro Well Drilling Planner, a tool that

supports the user in the design, execution and personnel training for ultra-

deep water petroleum well construction. Its scope spans from oil platform

selection to marine terrain editing and ocean waves rendering, as well as the

definition of the entire well drilling process. Vis-Petro is also built with a

flexible architecture, allowing the easy exchange among different graphical

libraries or virtual reality interfaces.

1. Introduction

Well planning and design is a very promising application area for Virtual Reality (VR),

since several integrated and interactive three-dimensional (3D) models are involved in

the process and should be precisely and realistically simulated, in order to make possible

their use by petroleum engineers.

This paper presents a computational tool, called Vis-Petro, which supports the

design, execution and personnel training for petroleum well construction, in the scenario

of submarine completion. The tool is composed of a set of modules, namely a platform

objects library, a terrain surface editing module, a reservoir editing module and a

module to define well design project, including pull points, equipments, tools and fluids.

The well design process is a very complex one. Nevertheless, there are already

some solutions for the definition of well paths [1]. The Vis-Petro tool implements a well

path definition model based on circle arcs and straight-line segments. In addition, in

order to enhance the visual quality of the application, the ocean surface simulates deep-

water ocean waves using a composition of well-known techniques [2] [3] [4]. Terrain

editing is also possible by the use of editing points together with adapted image

smoothing algorithms.

This paper intends to describe each of the implemented techniques as well as

highlight the relevance of such tool in the area of oil well planning. Section 2 explains

some basic concepts of the oil well drilling process. Section 3 reviews the current

research being carried out in oil well planning and VR. Section 4 presents the Vis-Petro

Drilling Planner tool, with its 3D interface and flexible editing. Section 5 details the

achieved results until the present moment. Lastly, section 6 gives the conclusion.

2. Well Planning Background

The process of planning an oil well demands large investments on account of the highly

specialized personnel and the specific tools needed for each different drilling situation

[5]. In order to reduce the risk of such an investment, a planning stage is of the utmost

importance. It not only reduces risks of not finding any oil, but it also optimizes the

drilling process, decreasing drilling costs, and therefore maximizing the final well profit.

By demanding the proper analysis of soil and terrain, water streams and sea tides

as well as personnel with experience in handling different types of drill, its bits, drilling

fluids, casings, tubes and rigs, well planning is one of the most complex engineering

activities. Therefore, a team of distinct experts in the areas of Geology, Oceanography,

Mechanics, Electronics, business management and others is gathered to accomplish it.

Each one of them contributes to the analysis and definition of the well drilling plan with

his distinct perspective, leading to a highly optimized well plan.

The process of well drilling is also non-trivial. Many are the variables involved,

most of them requiring the constant monitoring by this specialized team. Tools must be

constantly replaced and renewed and equipments must be positioned with temporal and

geographic precision to guarantee optimal drilling at minimum cost.

Before the well planning process itself starts, a data processing stage is

necessary. The user will use Vis-Petro to gather, compile, analyse, correct and model

previously collected data information about the ocean, its behaviour, the underlying

terrain layers and reservoir geologic constitutions. The result of this stage is a complete

3D model with all the necessary geographic information of the target region.

After processing the raw geo-data, the first step in the definition of the well

drilling plan is the selection of the oil platform, more generally called as the Marine

Perforation Unit (MPU), from which the oil well will be drilled and explored. This

choice will be based on the water and terrain conditions of the sea region.

Once this has been done, the well head, to be placed on the terrain surface, will

be selected together with the riser. The former works as a faucet and controls the oil

flow from the soil while the latter works as a hose, which must resist to the sea region

streams and tides, and also transport the oil to the MPU.

The third and last step is the definition of the well path through the different

terrain layers. That path will connect the oil reservoir to the well head. Associated with

each well section, there will be a set of tools and equipments that will make viable their

perforation. The process will be accomplished with the aid of immersive VR systems

such as CAVEs (Cave Automatic Virtual Environments) and special 3D visualization

rooms. Only after this well path is finished, re-evaluated and approved the drilling

activities will start.

This sequence of operations would never reach the current level of complexity

and perfection without the aid of computers. They are used in every drilling stage, from

the topographic mapping to the real-time control of the drill. Currently, few globally

renowned companies, such as Landmark©

[6], control the well drilling software market.

They provide software packets with programs to be used along the entire drilling

process. Nevertheless, they are still incapable of coping with all the nuances in the

drilling activity, mostly when it comes to specific well drilling scenarios, such as ultra-

deep water oil exploration.

3. Related Work

With regard to VR, research has enormously advanced, enhancing data visualization and

interaction during well planning.

The BP Centre has collaborated with innovative research in the area of data

visualization, drilling planning and geologic interpretation. Kenny Gruchalla has

proposed a well path planning system based in pull points [7]. These pull points are

inserted below the marine terrain surface with specific orientations and positions.

Sequentially connected by circle arcs and straight-line segments, these pull points define

the drill path connecting the well head to the reservoir, crossing through different soil

layers.

To provide realism to the virtual environment containing the well model, natural

phenomena must be simulated. These may include the simulation of ocean waves and

tides, sea streams, marine terrain, gravity and soil layers properties. Depending on the

complexity of the model, such simulations may become very realistic to the detriment of

performance. It is important, therefore, that the simulation model has only the degree of

realism required by the user, neither more nor less.

There are many models currently proposed for ocean waves simulation and

rendering. Hinsinger et al. [2] have proposed a marine wave simulation model to

represent a calm ocean. They represent waves far from coast and may be interactively

viewed by the user with constant performance guarantee. Ts’o et al. [3] have proposed a

rendering algorithm based on texture maps and the Fresnel Law of Reflection to

simulate wave refraction and reflection. It is a more efficient substitution to ray-tracing

techniques, although it is also less precise. Thon et al. [4] have proposed a filter

representing the wave spectrum generated by wind, which is applied to the automatic

wave trains generation. These are the basic techniques composing the Vis-Petro wave

model.

With regard to terrain editing and generation, many are the techniques and tools

available for automatic terrain creation [8]. Nevertheless, none of the terrain editing

techniques found which can be applied directly to the 3D model seemed adequate to

Vis-Petro edition necessities. Most of them do not allow editing terrain regions, but only

regenerating the entire surface. Moreover, the edition of reservoirs requires a different

approach since it is not restricted to a horizontal surface.

4. Vis-Petro: a Well Drilling Planner

The Vis-Petro tool aims to facilitate the process of well planning for petroleum

engineers. It has been architected with the purpose of being used in different platforms

and supporting a diverse number of interfaces such as desktop PC monitors, Head-

Mounted Displays (HMDs) and CAVEs. To achieve such flexibility, the architecture of

Vis-Petro has an independent conceptual model linked to a set of exchangeable 2D and

3D graphical and input interfaces.

The application’s 3D models are rendered by the OGRE (Object-oriented

Graphics Rendering Engine) engine while the WIMP (Windows, Icons, Menus and

Pointing devices) interface is rendered by a two-dimensional (2D) GUI (Graphical User

Interface) library, such as CEGUI [10].

OGRE is an open-source general-purpose graphical rendering engine. It works

with most operating systems and video cards. OGRE’s native interface has been written

in C++, using advanced object-oriented concepts such as inheritance and polymorphism.

The architecture of Vis-Petro is divided in three main modules. The first one

manages the interaction with the user, the second contains the entire conceptual model

and the third manages internal operations. Figure 1 shows an interaction diagram of its

main architectural components.

Figure 1. General Vis-Petro architecture.

The components 2DModule, StateManager and State compose the

interaction module. 2DModule controls application menus via interaction commands

and is implemented by an external 2D GUI. The second component, State, assumes

the possible editing states of the system. It is responsible for applying the interaction

captured from the user in the 3D world to the conceptual model. There is a specific state

for MPU editing, as well as for terrain, well head, well path, fluid, tools and equipment

editing. Each state contains general sub states, such as selecting, editing or removing an

object from the scene. The third and last component, StateManager, is the most

important one of Vis-Petro. It manages all the possible states of the system. It also

connects all other main components.

The ConceptualModel is the second module and manages the attributes and

behaviours of the conceptual objects such as equipments, MPUs, pull points, reservoirs,

sea, terrain, tools and wells. It is the only access point to those conceptual objects by the

interface components, hence enabling the flexible adaptation of the conceptual model to

other graphical interfaces.

FileManager and ViewpointManager constitute the third and last

module and execute extra-internal operations. Vis-Petro allows the user to create model

libraries from project models for future reuse. For that purpose, FileManager is

responsible for loading and saving models, as well as full projects. In addition,

ViewpointManager is responsible for controlling the user viewpoints, saving and

restoring them. Viewpoints are associated to the states that were active in the moment of

their insertion, even though they are accessible from other states.

To exemplify the interaction among those main components, the steps necessary

to add a fluid to a well path are described in sequence and shown in Figure 2: a) The

user sets the application to the fluid editing state on the application menu (2DModule);

b) The user selects the type of fluid to be applied to the current well section 2DModule

activates the fluid insertion sub state via StateManager, obtains a new fluid of the

selected type from FileManager and forwards it to StateManager, which

forwards the fluid information to State; c) The user selects a well in the 3D scene,

ConceptualModel returns it to State; d) The user selects the fluid insertion points

on the well path and State adds the new fluid to ConceptualModel.

Figure 2. Adding a fluid to a well path.

Similar processes occur for each possible interaction in Vis-Petro. It is important

to notice, in this example, the intermediation of StateManager between the active

state in the 3D scene and the 2D interface.

Regarding user movement, two different navigation modes are possible. The first

one allows pitching and yawing, as well as translation on the user’s head direction and

its perpendicular planes. The second also allows yawing and pitching, but the translation

movement only occurs on planes parallel or perpendicular to the sea surface.

5. Methodological Results

Vis-Petro has a prototype developed with the OpenGL render library. It consists of a

well path creation model with pull-point edition and models for ocean waves simulation

and marine terrain surface. There is also an interface prototype developed in

Macromedia Flash©

to demonstrate Vis-Petro functionalities. The architecture of Vis-

Petro has also been defined and its implementation is currently being carried out.

With regard to interface devices, Vis-Petro’s prototype has already been

successfully adapted to some advanced devices, such as six-degrees-of-freedom (6DOF)

gloves and stereoscopic desktop glasses. In addition, it has been evaluated in both

Windows and Linux operational systems.

The following sections will explain in detail the underlying mechanisms for well

path creation, ocean waves simulation, marine terrain surface construction, and user

interaction. The results achieved, until the current moment, are presented, as well as the

explanation about the methodologies successfully implemented.

5.1. Well Path Creation Model

The module responsible for the construction of the well path connecting the well head to

the oil reservoir was implemented based on the technique presented by Gruchalla [1].

The well path consists of a set of adjustable arrows, denominated pull points, which are

automatically connected by a sequence of circle arcs and straight-line segments.

The model implemented allows real-time adjustment of the well path by simply

translating or rotating the pull points. The current version places all pull points on a

plane perpendicular to the sea surface. Two kinds of well path editing techniques are

possible. In the first type, the user manipulates each pull point. The pull points may

rotate around the plane normal, with the latter positioned in the pull point centre, and

translate in the plane surface. Three types of connections between two pull points are

possible: a straight-line segment, a C-shaped curve followed or not, or preceded, by a

straight-line segment, and an S-shaped curve having or not a straight-line segment in its

middle. They are respectively presented in Figure 3a.

(a) (b)

Figure 3. (a)Types of connections between pull points and (b) a Vis-Petro pull point well path example.

In the second type of edition, the plane containing the pull points may be rotated

around the normal of the well head, changing the azimuth angle of the entire well path.

The well head, together with that path, may also be translated. Therefore, a well head

may be placed anywhere in the surface of the terrain and may be connected, through the

well path, to any point in a reservoir’s surface.

In the current version, azimuth is not implemented, that is, well path creation is

restricted to a plane. Figure 3b shows an example of well path. The pull points are

presented as cones, the first one representing the well head.

The well path model is very flexible in terms of well editing. It is built during

pull point insertion. As soon as a new pull point is inserted, the connection from the last

point to it is generated. The pull points position or orientation and the well path are

updated in real-time. The distinction between types of sections helps eliminate

redundant intermediate points in straight well sections. Nevertheless, the user himself

must adjust the pull points to ensure they do not collide with other wells in production

or with undesirable geological structures.

5.2. Ocean Waves Simulation Model

Graphical simulation of natural phenomena, depending on the realism desired, may

become a non-trivial and computationally expensive task. Nevertheless, the simulation’s

visual and physical quality must satisfy users, who would like to feel immersed in the

virtual world as well as accomplish interaction tasks consistently. To enhance Vis-Petro

realism, ocean waves simulation was implemented.

Currently, a number of ocean waves models are proposed. The present work was

based on Hinsinger et al., Ts’o et al. and Thon et al., for modelling, rendering and

animation [2][3][4]. The relation between each of these models and Vis-Petro

implementation is presented below.

5.2.1 Wave Physical Modelling

Following Hinsinger et al. [2], Vis-Petro’s wave model was based on Gerstner’s

swell model, which describes the behaviour of ocean particles. Wave trains were

simulated with the shape of trochoids. The Gerstner’s model sets each water particle to

move in circles, whose radii are equal to the amplitude of the wave passing through that

particle, and around the location at rest.

However, the combination of several different waves was needed to provide

greater realism. Therefore, the sum of wave train functions generated the final shape of

the ocean surface. In order to generate a calm sea spectrum, automatic wave trains

generation was used, similar to the work of Thon et al [4]. The model Thon et al was

also used to define angular velocity and wave amplitude. The remaining wave

parameters, wavelength and wave direction, were defined manually.

An anti-aliasing technique was also used to prevent elevations in regions where

distance would make them imperceptible to the user. When the sum of heights for a

location at rest is calculated, the wave trains whose wavelengths do not exceed the

distance between that location and its nearest neighbour are discarded. This technique

improves system performance.

The definition of the particles’ location at rest on the ocean surface was achieved

by applying the adaptive scheme proposed by Hinsinger et al [2]. Accordingly, it

projected a mesh from the screen to the horizontal plan cz = (where the constant ℜ∈c )

to obtain the locations at rest of the Gerstner’s model. Those locations were used as

input to the wave trains equation. In this model, the sampling by projection ends up

having the Level of Detail (LOD) technique intrinsically implemented. Moreover, as the

ocean surface mesh resolution is constant, constant execution time is also guaranteed,

independently of the ocean’s region being visualized. In this work, the ocean particles

were uniformly distributed over each axis.

5.2.2 Results

The result of the Vis-Petro’s ocean waves simulation is presented in Figure 4.

The results obtained from this model were quite realistic, but implied in a perceptible

frame rate decrease. The uniform distribution of the mesh vertices ended up generating a

high number of points, which slightly hindered optimal performance. The simulation of

ocean waves is being integrated into the architecture of Vis-Petro as one of the scene

components inside the conceptual model with a special connection to the 3D interface.

Figure 4. Results of Vis-Petro ocean waves simulation model.

5.3. Marine Terrain Surface

The Vis-Petro application allows the user to edit the marine terrain surface, creating a

terrain profile composed of peaks and depressions, resulting in realistic models that do

not restrict user solely to standard surfaces, such as planes. The marine terrain surface is

constituted by a height map which may be edited to model those peaks or depressions.

The marine terrain surface has all cells divided in their middle to achieve a

higher LOD in the mesh. Each intersection point between lines and columns, henceforth

called editing points has its height editable.

Every point contained in the editing point’s actuation area will have its height

modified according to an editing point height-based function.

To create a peak or a depression when the user changes the height of an editing

point EP, a function Height is applied to the points contained in its actuation area. The

Equation Erro! A origem da referência não foi encontrada. is responsible for

producing such an effect, creating a peak or a depression depending on whether the

value of the editing point height is positive or negative, respectively.

( ) ( ) ( )pp

p

p

p

p hzzarea

hxx

area

hzxHeight +−

−+−

−=

2

2

2

2, ( 1 )

x is the x coordinate of a point contained in the actuation area of EP, z is the z

coordinate of a point contained in the actuation area of EP, px is the x coordinate of the

editing point, pz is the z coordinate of the editing point, ph is the editing point’s height,

area is the editing point’s actuation area and pHeight is the function which calculates

the height of a point ( )zx, according to the editing point ( )pp zx , height.

Another impact due to the modification of the editing point height occurs when

the actuation areas of more than one point intersect. The calculus of the affected points’

heights may be given by three different functions: the sum of the heights resultant from

the actuation of each editing point (additive interpolation), the superposition of the

maximum heights generated by the actuation of each editing point (maximum restrictive

interpolation) or the superposition of the minimum heights generated by the actuation of

each editing point (minimum restrictive interpolation).

To explain each of the above interpolations, given a 3D surface with dimensions

50x50 and two editing points: ( ){ }15,50,25,251 =−==== areahzxP and

( ){ }5,25,25,252 ===== areahzxP , Figure 5 shows the achieved results with the

additive, maximum restrictive and minimum restrictive interpolations of the points 1P

and 2P , respectively.

(a) (b) (c)

Figure 5. (a) Additive interpolation, (b) maximum restrictive interpolation and (c) minimum restrictive interpolation without smoothing.

However, by having a closer look in the example above, it is clearly visible that

the results do not look very realistic and other techniques must be taken into

consideration. To avoid those sharp contours on the above surfaces, a smooth algorithm

was applied in the entire region affected by the editing points.

Box Filtering [11], known as a method to reduce the intensity variation between

pixels of an image, was here applied to smooth the 3D surface and reduce the height

variation between points. Although it is not being applied on an image, the surface

heights consistently substitute the image’s pixels. Figure 5b is an example of a surface

area that should be smoothed.

Figure 6. Maximum restrictive interpolation with smoothing.

The Box Filtering algorithm was applied on the bounding area resultant from the

union of the actuation areas of the editing points that intersected. The result, shown in

Figure 6, makes explicit the benefits of smoothing in the terrain surface creation.

Figure 7 shows a complete marine terrain surface edited from a plane as a result

of user interaction, which is currently being carried out via a configuration file. Such

configurations consist in setting the dimensions of surface, the interpolation models for

each editing point and the positions and heights of the editing points.

Figure 7. A marine terrain surface.

5.4. Interaction

The interaction system of the Vis-Petro application was developed with the necessary

flexibility to be applied to a variety of interface types. These include from desktop to

immersive interfaces, such as HMDs and CAVEs. The interface will be adjusted to each

new set of peripherals, having slight modifications in the set of commands, but

maintaining the core interaction system.

This interaction core is composed by navigation, selection, manipulation,

edition, insertion, removal and viewpoint commands. They are described in detail

below. These commands may be divided in two distinct interaction categories: 2D

interface commands and direct 3D commands.

The first one contains commands made via the 2D lateral menu, illustrated in

Figure 8. The selection of different application states, the insertion of models by

dragging them from this menu to the 3D scene and the configuration of models

parameters are interaction activities only possible through the use of such menu.

On the other hand, the direct commands are originated from an input device and

are directly applied to the 3D scene. These include commands for navigation, selection

and manipulation of scene objects, as well as pull points themselves.

Nevertheless, many are the commands that may be applied by both interaction

techniques, due to the existence of convenient shortcuts. This is the case of viewpoint

manipulation, save and open project, 3D object translation, rotation, removal and state

modification commands.

While in desktop interfaces most commands may be done via the lateral menu

without problems, in immersive interfaces, commands will have to be done directly in

the application in order not to affect user immersion.

A prototype of the desktop interface is shown in Figure 8. A study on which

interface library will better comply with the necessities of the application is currently

being carried out. It is composed of an application menu with commands for

construction of an oil well project. This menu allows the user to open and save projects,

models and libraries, insert, restore and delete viewpoints from the model, insert, edit

and delete objects from the scene, navigate (change the user point of view) in the scene.

Moreover, not only are they available by accessing the lateral menu, but also by using

keyboard shortcuts. An interface prototype for Vis-Petro may be found at

http://www.gprt.ufpe.br/~grvm/vispetro.

Figure 8. Interface prototype for desktop based application version.

6. Conclusions and Future Work

The analysis of Vis-Petro functionalities shows that each of them has their benefits for

each problem they aim to solve. Their analysis, however, posed different perspectives in

the evaluation of the well planning application development. It is believed that the

thorough study and further improvement of these techniques will contribute to the

definition of a model capable of better coping with the problems arisen during well

planning.

Although the sea model mesh points were uniformly distributed, other

approaches, in which the density of points would vary according to LOD levels, could

have been used. They could have made possible the mesh resolution reduction without

loss of visual quality. As long as waves are only a visual enhancement to the model, not

having their physical properties any effect over well planning, such optimisations may

be applied in future versions without loss simulation precision. However, sea behaviour

interaction, such as collision, must also be considered in future tool versions.

With regard to the well path, much as the model is consistent, it is also missing

the addition of complementary functionality. The manipulation of the azimuth angle will

greatly expand the user interaction possibilities. Apart from that, another interesting

possible feature would be the automatic collision detection and avoidance among wells

in a region.

Regarding marine terrain edition, despite Vis-Petro’s capability in dealing with

such matter effectively much still must be done. Better interpolation and smoothing

techniques, with approaches specific to the terrain manipulation problem, must be

designed. Nevertheless, the study and test of techniques from other areas of knowledge,

such as image processing, have proved to be a quite interesting initial approach.

The edition of reservoirs is currently under development and will be based on

marine terrain edition, using edition points and the edition point’s normal to indicate the

peak or depression growth direction.

Currently, the data used by Vis-Petro is still simulated. The future access to real

geographic data will greatly help in the validation of the implemented techniques.

Lastly, although some advanced input and output devices have already been

tested, others will be analyzed and interaction with each one of them will be evaluated.

Much as Vis-Petro has progressed, it still needs further improvements and

complements. Other modules are under development to increase the toolset available

and enable Vis-Petro to take part in a greater number of well planning activities.

Nevertheless, its innovative interface, integration with the entire ultra-deep water

drilling process and flexibility to different interfaces and platforms makes of Vis-Petro a

singular tool for off-shore well planning.

7. Acknowledgements

The authors are grateful to CNPq, which partially funded this research.

References

[1] K. Gruchalla, Immersive Well Path Planning: The Added Value of Interactive Immersive

Visualization, master's thesis, BP Centre for Visualization, University of Colorado at Boulder, 2003.

[2] D. Hinsinger, F. Neyret, and M.P. Cani, “Interactive Animation of Ocean Waves”, ACM

SIGGRAPH/Eurographics SCA, Saarbruecken, Germany, 2002, pp. 161-166.

[3] P.Y. Ts’o, and B.A. Barsky, “Modeling and Rendering Waves: Wave-Tracing Using Beta-Splines

and Reflective and Refractive Texture Mapping”, ACM TOG, vol. 6, issue 3, ACM Press, 1987, pp.

191-214.

[4] S. Thon, J.-M. Dischler, and D. Chazanfarpour, “Ocean Waves Synthesis using a Spectrum-based

Turbulence Function”, CGI, Geneva, Switzerland, 2000, pp. 65.

[5] J.E. Thomas, Fundamentos da Engenharia de Petróleo, 2nd

edition, Editora Interciência, Rio de

Janeiro, Brazil, 2001.

[6] Landmark Graphics Corporation. Available: Landmark Graphics Corporation Site. URL:

http://www.lgc.com/, visited on March 2006.

[7] K. Gruchalla, “Immersive Well Path Planning: Investigating the added value of immersion”, IEEE VR

Conference, Boulder, Colorado, USA, 2004.

[8] C. Hormann, Terrain Simulation Links. Available: C. Hormann Site. URL:

http://www.imagico.de/terlinks.html, visited on January 2006.

[9] D.A. Bowman, E. Kruijff, J.J. LaViola, and I. Poupyrev, 3D User Interfaces: Theory and Practice,

Addison Wesley Longman Publishing Co. Inc., Redwood City, CA, USA, 2004.

[10] OGRE 3D: Open Source Graphics Engine. Available: The OGRE Team Site. URL:

http://www.ogre3d.org/, visited on January 2006.

[11] R.C. Gonzalez, and R.E. Woods, Processamento de Imagens Digitais. Editora Edgard Blucher Ltda,

São Paulo, SP, Brazil, 2000.